Add doc_url field to match its recent addition in pprof proto. #588
Conversation
github.com/google/pprof/pull/888 added this field in pprof proto, this change adds it to the OTel profiling proto to match the change.
|
@florianl @jhalliday FYI |
opentelemetry/proto/profiles/v1experimental/pprofextended.proto
Outdated
Show resolved
Hide resolved
| // | ||
| // The URL may be missing if the profile was generated by older code or code | ||
| // that did not bother to supply a link. | ||
| int64 doc_url = 19; // Index into string table. |
There was a problem hiding this comment.
With ScopeProfiles.schema_url the URL that describes the provided profile exists already. Where do you see the differences between the two, so that this additional field is required?
There was a problem hiding this comment.
schema_url points to https://opentelemetry.io/docs/specs/otel/schemas/#schema-url file which is not user docs. The doc_url here is for user-level documentation. I would also expect that the schema URL would be the same for many profiles as it corresponds to the version of the data schema or something like that rather than the nature of the data (CPU profile vs heap profile vs contention profile).
There was a problem hiding this comment.
Since this should be an optional field, what do you think about using an attribute?
On a similar note, we had a discussion in the last SIG about the boolean fields in Mapping and the consensus was to drop them from the proto and turn them into attributes.
There was a problem hiding this comment.
I think the doc_url can be an attribute. I'll change this.
For the has_* fields I'm less sure. I added an agenda topic to the SIG notes doc with some thoughts on this.
Co-authored-by: Florian Lehner <florianl@users.noreply.github.com>
google/pprof/pull/888 added this field in pprof proto, this change adds it to the OTel profiling proto to match the change.